The intensive care unit (ICU) is the highest mortality place in any hospital and is one of the sites in which patient safety deficiencies are most common. For these reasons, the JCAHO, the NQF, and the Leapfrog Group are considering requiring or recommending ICU performance reporting. Despite the evidence of variation in care and outcome and the interest of policymakers, little is known about how to evaluate ICU performance. There have been no prior attempts to determine whether, using existing risk adjustment models, an excess of observed over expected mortality is an indicator of patient safety deficiencies in processes of care of simply reflects unmeasured clinical determinants of severity. It also is unknown whether structural variable that are associated the lower risk-adjusted mortality are actually associated with better process of reflect the influence of unmeasured confounders. In this project, we will build on existing research about evaluation of structure, process, and outcome in the ICU with the following objectives: 1) to determine the extent to which mortality reports identify hospital with high rates of errors in processes of care, 2) to evaluate the relationship between structure and processes of care and 3) to assess the extent to which provision of patient safety performance data can stimulate improvement. The central hypotheses underlying this work are that 1) risk adjusted mortality rates and structural variables are valid indicators of deficiencies in processes of care that pose patient safety threats, and 2) that these indicators can be used to drive improvements in outcomes. We will build on the existing 46 hospital CALICO network-members of which currently collect the detailed clinical data required to calculate all extant ICU mortality risk-adjustment models on their ICU patients-to create a more complete picture of ICU care than ever previously available for a hospital sample of this size. We will accomplish the goals of this project by performing chart abstractions to assess processes of care at CALICO hospitals and ICU surveys to assess structure, than linking existing mortality data to processes and the structural survey data to processes. We will then give participating hospitals feedback about their structure, processes, and outcomes and repeat our performance evaluation to assess change. The lessons of this project will help policy makers decide what how to measure and improve patient safety in the ICU.